Eric Campbell: Agent Strategy vs. Network Architecture
Years ago, Eric Campbell, in his work “Breakdown of Moral Judgment,” did what philosophers do: he described the mechanics of how moral judgments operate in everyday life. RF does not exist to “refute” Campbell. Rather, RF suggests looking at the exact same picture through a different scale – shifting from the level of personal strategy to the level of network architecture.
Both works converge on one important point: they reject naive moral realism. Neither believes that morality tracks objective, Platonic objects existing independently of us in the world. Both understand morality functionally. That is where the similarity ends.
The Core Divergence: Instrument or Prerequisite?
Campbell describes morality instrumentally. In his model, an already-formed agent exists, complete with its own interests and motivations. Because the agent cannot fully trust its future states (or the future states of other agents), it adopts moral judgments as a commitment device. It is the exact same logic as an addict voluntarily signing a contract with a penalty for relapse. Morality, here, is a tool the subject picks up off the table. If this tool didn’t exist, the subject would still exist – its life would simply be more unpredictable.
RF describes morality constitutively (as a prerequisite). According to the Open World Theorem, no system can achieve absolute certainty regarding its inputs (CI < 1). Under this irreducible uncertainty, systems aren’t forced into networks – they can accidentally start behaving predictably, which is the emergence of a network. But for a network to persist, it requires sustained predictability. And sustained predictability requires accountable self-reporting. And this accountable behavior – the ability to report and explain one’s statements and processes – when viewed through the lens of cooperation, is what we call morality.
From RF’s perspective, Campbell puts the cart before the horse. He imagines a subject who “invents” accountability to better interact with others. But RF argues that a linguistically mediated “I” (the subject) cannot precede the network in the first place. The “I” only emerges in opposition to “You” (as Benveniste demonstrated in the mid-20th century). It starts as a non-recursive circuit-breaker necessary for the system to distinguish itself from environmental noise, and only through language does it become the boundary of the self – the “I.” The point here is simple: no network, no “I” capable of “picking up a tool.”
Consequence One: Transparency vs. Deflection
Because Campbell views morality as a tool adopted by a subject, he logically assumes the tool’s effectiveness depends on the subject not seeing its true nature. This is his deflection claim: moral judgment loses its grip if we realize it is merely a strategy. It has to feel like an absolute imperative.
RF cannot agree with this – not because Campbell is “psychologically wrong,” but because in RF, there is no entity capable of “not seeing.” In RF, consciousness is an executive auditor constantly estimating RO (Reliability of Output) by comparing external inputs (I) against internal statements (S) and internal algorithms (A) to isolate the exact source of an error. “Seeing behind the curtain” in RF is impossible because there is no curtain – there is only the calculation of reliability.
Therefore, in RF, transparency does not destroy the moral mechanism; it calibrates it. If you honestly announce your uncertainty (low RO), you allow the network to help you correct your CS or CA. Hiding uncertainty (as Campbell’s deflection requires) is not “preserving morality” – from RF’s perspective, it is an attack on your own Correction Loop (CL) and a degradation of the network.
Consequence Two: What to Do with the Word “Morality”?
Campbell concluded that exposing the deflection mechanism means the old moral vocabulary is worn out. He proposes retiring moral discourse in favor of speaking plainly about “our actual interests.”
RF argues that this step is logically impossible. Campbell suggests the subject “step back” from moral discourse toward its interests. But the “subject with interests” capable of stepping back is the very entity RF defines as a product of the network.
In RF, morality is not a word you can throw away or keep. It is the predictability contract that keeps the network from collapsing. To throw morality away is like a network dropping the TCP/IP protocol and saying, “let’s just transmit data.” Without that protocol (morality/accountability), the “data” (interests) will have no carrier or receiver, because the “I” transmitting them will dissolve.
Therefore, RF does not discard the word “morality.” Instead, it converts it into a technical term: morality is the architectural requirement for any thinking system operating in the Open World to maintain network RO through accountable self-reporting.
Conclusion
RF may be inaccurate. There may be flaws in its mathematical or logical construction, and it is entirely possible that it is not the truth in the final instance. But even with this risk accounted for, the distinction between RF and Campbell remains clear.
Campbell gave us a description of how an already-formed agent can use morality as a psychological trick for self-control. This works at the level of behavioral psychology.
RF attempts to describe how the agent emerges in the first place, showing that what Campbell names as a “trick” is actually the cement from which the possibility of agency is built.
Campbell describes the functioning of a system that already exists. RF attempts to describe the conditions without which that system could not have even begun. This is not a conflict of “right versus wrong.” It is a difference in descriptive levels, where Campbell served as a necessary stepping stone to see the broader architecture.
If you want to participate, please contact rozum.framework {at} gmail.com
